The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill instructs the agent to clone the repository from https://github.com/mcncarl/agent-memory-vault.git to the local environment.\n- [REMOTE_CODE_EXECUTION]: The skill executes Python scripts downloaded from the external repository, such as scripts/install-posix.py, to perform installation and configuration tasks.\n- [COMMAND_EXECUTION]: Various local commands are used, including python3 for script execution, git for repository management, and a custom memoryctl tool for interacting with the vault.\n- [PERSISTENCE]: The skill contains instructions to create macOS LaunchAgents or Windows Task Scheduler tasks for automated memory maintenance, which allows for persistent background execution.\n- [INDIRECT_PROMPT_INJECTION]: The skill ingests and processes Markdown files that may contain instructions targeting the agent.\n
Ingestion points: External data is read into the agent context through the memoryctl search and retrieve commands from the vault files as described in references/daily-reads.md.\n
Boundary markers: The skill uses stdin JSON and schema versioning for structured data passing, but does not implement explicit delimiters to isolate retrieved memory content from agent instructions.\n
Capability inventory: The system is capable of executing local Python scripts, writing to the file system, and establishing persistent tasks.\n
Sanitization: Safety rules in SKILL.md and leak-checking procedures in references/template-development.md are used to prevent sensitive data exposure, but do not specifically sanitize against prompt injection.

yichen-agent-memory